Active garment recognition and target grasping point detection using deep learning. (February 2018)

• Record Type: Journal Article
• Authors: Corona, Enric; Alenyà, Guillem; Gabas, Antonio; Torras, Carme
• Is Part Of: Pattern recognition. Volume 74(2018:Feb.)
• Page Start: 629
• Page End: 641
• Publication Date: 2018-02
• Subjects: Garment classification -- Garment grasping -- Deep learning -- Depth images; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2017.09.042
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: We propose an algorithm that first, identifies the type of the garment and second, performs a search of the two grasping points that allow a robot to bring the garment to a known pose. Using Maya, we generate a database of depth images from simulated garments. The whole process is automatized by a code we make public. We combine depth images from real garments with simulated data, to train a Convolutional Neural Network that significantly improves state of the art results in cloth recognition. To detect the visibility and Cartesian location of the reference points, we use two more Convolutional Neural Networks per garment. The garment manipulation we propose differs from the classical approach based on re-grasping of the lowest hanging parts. Abstract: Identification and bi-manual handling of deformable objects, like textiles, is one of the most challenging tasks in the field of industrial and service robotics. Their unpredictable shape and pose makes it very difficult to identify the type of garment and locate the most relevant parts that can be used for grasping. In this paper, we propose an algorithm that first, identifies the type of garment and second, performs a search of the two grasping points that allow a robot to bring the garment to a known pose. We show that using an active search strategy it is possible to grasp a garment directly from predefined grasping points, as opposed to the usual approach based on multiple re-graspings of the lowest hanging parts. Our approach uses a hierarchy of three Convolutional Neural Networks (CNNs) with different levels of specialization, trained both with synthetic and real images. The results obtained in the three steps (recognition, first grasping point, second grasping point) are promising. Experiments with real robots show that most of the errors are due to unsuccessful grasps and not to the localization of the grasping points, thus a more robust grasping strategy is required.